Home

iCAD Overview

iCAD is an audio processing pipeline designed for radio listening applications. It processes audio files by detecting tones, transcribing content, archiving data, and enabling playback and alerting functionalities.

Supported Radio Applications

Currently supported:

• Trunk Recorder (via MQTT or Upload Script)

Planned support:

• RTL Airband
• ProScan
• Direct Audio Input

Key Components and Features

Audio Consumer

Responsible for processing incoming audio and metadata:

• Extracts tone segments and
• Sends Audio for transcription.
• Archives both audio and metadata.
• Uploads content to various listening services.
• Triggers alerts through iCAD Alerting based on specific conditions.

Tone Detection

Identifies different types of tones within audio files:

• Two Tone
• Long Tones
• Hi/Low (Warble) Tones

Transcription

Utilizes Faster Whisper for audio-to-text conversion:

• Dynamically configures transcription settings per audio file.
• Excludes tone segments during transcription for improved accuracy.
• Associates transcribed text with corresponding metadata segments.

Archiving

Stores files on accessible servers:

• Local webserver
• Amazon AWS S3
• Google Cloud

Listening Services

Enables audio file access through various platforms:

• OpenMHZ
• Broadcastify Calls
• iCAD Player
• RDIO (Supports multiple instances)

iCAD Alerting

Generates alerts based on detected tones and transcribed text:

• Configurable triggers for tone and transcription filters.
• Notification methods include email, Pushover, Facebook, Telegram, and webhooks (supports multiple webhooks per system and per trigger).